{
  "markdown": "# Claude Code Global Skills Repository\n\nCentralized skills and commands for Claude Code, symlinked into projects via `enable-skills.sh`. All projects share and auto-update from this single source.\n\n## Table of Contents\n\n- [Setup](#setup)\n- [Skills](#skills)\n  - [Analysis](#analysis)\n  - [Bioinformatics](#bioinformatics)\n  - [Claude Meta](#claude-meta) (essential — auto-included)\n  - [Collaboration](#collaboration)\n  - [Databases](#databases)\n  - [Galaxy](#galaxy)\n  - [Packaging](#packaging)\n  - [Project Management](#project-management) (essential — auto-included)\n  - [VGP](#vgp)\n- [Commands](#commands)\n- [Hooks](#hooks)\n- [Adding New Skills](#adding-new-skills)\n\n## Setup\n\n```bash\n# Set environment variable (add to ~/.zshrc)\nexport CLAUDE_METADATA=\"$HOME/Workdir/claude_global\"\n\n# Enable skills in a project\ncd /path/to/your/project\nbash $CLAUDE_METADATA/enable-skills.sh\n```\n\nThe setup script detects project type and suggests appropriate skills:\n\n| Project Type | Suggested Skills |\n|---|---|\n| **Analysis/Research** | All `analysis/*`, `bioinformatics/*`, `databases/*`, `collaboration/hackmd`, `collaboration/project-sharing` |\n| **Development** | `packaging/conda-recipe` |\n| **Bioinformatics** | All `galaxy/*` |\n\nAll `claude-meta/*` and `project-management/*` skills are always included as essentials.\n\n## Skills\n\n### Analysis\n\n| Skill | Description |\n|---|---|\n| `data-analysis-patterns` | Data aggregation, recalculation vs reuse, category management, analytical accuracy. |\n| `data-visualization` | Publication-quality figures with matplotlib/seaborn. Journal-specific `.mplstyle` files (Nature/Science/Cell), colorblind-safe palettes, figure export helpers, Claude API image constraints. |\n| `documentation-organization` | Structure working files, prepare sharing packages, clean project layout. |\n| `jupyter-notebook` | Comprehensive notebook analyses with statistical rigor, outlier handling, publication-quality visualizations. |\n| `scientific-publication` | Iterative refinement of publication figures — systematic improvement, layout optimization. |\n\n### Bioinformatics\n\n| Skill | Description |\n|---|---|\n| `fundamentals` | SAM/BAM, AGP, sequencing technologies (Hi-C, HiFi, Illumina), quality metrics, alignment debugging. |\n| `phylogenetics` | Phylogenetic tree analysis, visualization, annotation management, iTOL troubleshooting. |\n| `visualization` | Publication-quality bioinformatics figures — phylogenetic trees, genome browsers, iTOL datasets. |\n\n### Claude Meta\n\nEssential skills defining how Claude works with you. **Auto-included in all projects.**\n\n| Skill | Description |\n|---|---|\n| `collaboration` | Team workflows — skill management, knowledge capture, version control. |\n| `documentation` | Session documentation — incremental summaries, fix reports, audit trails. |\n| `skill-management` | Creating, symlinking, updating, and organizing the centralized skill repository. |\n| `systematic-debugging` | 4-phase debugging: root cause → pattern analysis → hypothesis → fix. Anti-rationalization tables. |\n| `token-efficiency` | Token optimization — efficient file reading, command execution, model selection (Opus for learning, Sonnet for dev). |\n| `verification-before-completion` | Evidence-based completion claims. Run verification before claiming success. |\n\n### Collaboration\n\n| Skill | Description |\n|---|---|\n| `hackmd` | Slide presentations, embedded SVG diagrams, real-time collaborative editing. |\n| `project-sharing` | Prepare organized sharing packages at different levels (Summary/Reproducible/Full). |\n\n### Databases\n\n| Skill | Description |\n|---|---|\n| `bioservices` | Unified Python interface to 40+ bioinformatics services (UniProt, KEGG, ChEMBL, Reactome, PSICQUIC). Cross-database analysis and ID mapping. |\n| `gget` | Fast CLI/Python queries to 20+ databases — gene info, BLAST, AlphaFold, enrichment, single-cell, disease associations. |\n| `gnomad` | gnomAD GraphQL API — population allele frequencies, constraint scores (pLI, LOEUF), variant pathogenicity, ACMG criteria. |\n\n### Galaxy\n\n| Skill | Description |\n|---|---|\n| `automation` | BioBlend and Planemo — Galaxy API, workflow invocation, batch processing, dataset management. |\n| `tool-wrapping` | Galaxy tool wrapper XML, Planemo testing, best practices. |\n| `training-material` | GTN tutorial development — markdown syntax, special boxes, tool references, YAML front matter. |\n| `workflow-development` | Galaxy .ga workflows, IWC standards, testing, optimization. |\n\n### Packaging\n\n| Skill | Description |\n|---|---|\n| `conda-recipe` | Conda/bioconda recipe creation, linting, dependency management, build debugging. |\n\n### Project Management\n\nEssential skills. **Auto-included in all projects.**\n\n| Skill | Description |\n|---|---|\n| `data-backup` | Smart backups with project type detection. Rolling daily, compressed milestones, CHANGELOG. |\n| `folder-organization` | Project folder structure, file naming conventions, directory standards. |\n| `managing-environments` | Python venv and conda environment management. |\n| `obsidian` | Obsidian vault integration — notes, tasks, knowledge management, MOCs, CLI (1.12+). |\n\n### VGP\n\n| Skill | Description |\n|---|---|\n| `genomeark-aws` | GenomeArk AWS S3 bucket — VGP assemblies, QC data, species directories. |\n| `vgp-pipeline` | VGP assembly pipeline — Galaxy workflow selection, QC checkpoints, batch orchestration. |\n\n## Commands\n\n### Session Management\n\n| Command | Description |\n|---|---|\n| `/safe-exit` | End session with backup and Obsidian summary. |\n| `/safe-clear` | Save notes to Obsidian, update skills, clear context. |\n\n### Project Organization\n\n| Command | Description |\n|---|---|\n| `/backup` | Smart backup with skill-aware cleanup. |\n| `/consolidate-notes` | Consolidate session notes with AI-powered analysis. |\n| `/cleanup-project` | Remove working docs, condense verbose READMEs. |\n| `/deprecate-file` | Move files to deprecated/ with dependency handling. |\n| `/share-project` | Prepare sharing packages (Summary/Reproducible/Full). |\n\n### Navigation & Discovery\n\n| Command | Description |\n|---|---|\n| `/command-help` | Show help for any command. |\n| `/list-skills` | List all available skills with descriptions. |\n| `/read-manifest` | Smart session startup — load relevant context. |\n| `/generate-manifest` | Generate or update MANIFEST.md file inventories. |\n| `/update-manifest` | Quick-update MANIFEST.md preserving user content. |\n\n### Setup & Maintenance\n\n| Command | Description |\n|---|---|\n| `/setup-project` | Set up Claude Code skills for a new project. |\n| `/setup-environment` | Plan and set up Python venv or conda environment. |\n| `/sync-skills` | Sync project with global metadata — detect new skills/commands. |\n| `/update-skills` | Review session and suggest skill updates. |\n| `/update-notebook` | Notebook maintenance — figures, references, TOC, coherence. |\n\n### Planning\n\n| Command | Description |\n|---|---|\n| `/design-and-plan` | Brainstorm → design → plan → execute workflow for complex tasks. |\n| `/plan-interview` | Deep interactive Q&A about a plan, then rewrites it with your answers. |\n| `/plan-research-questions` | Parallel subagents research each open question in a plan. |\n| `/plan-integrate-questions` | Merge answered questions back into the plan body. |\n| `/plan-review` | Review an unimplemented plan for clarity, correctness, and feasibility. |\n| `/plan-debrief` | Compare implementation against a plan — find gaps and propose next steps. |\n| `/plan-summary` | Summarize next steps from a plan with repo verification. |\n| `/plan-clean` | Renumber phases/steps sequentially and update all cross-references. |\n\n### Code Quality\n\n| Command | Description |\n|---|---|\n| `/dedup` | Semantic duplication analysis with concrete refactoring recommendations. |\n| `/review-test` | Review tests for quality — challenge weak assertions, find dead tests. |\n| `/challenge-mocks` | Challenge Python mock/patch usage, suggest DI, fakes, and real alternatives. |\n\n### Git Workflow\n\n| Command | Description |\n|---|---|\n| `/decompose-for-rebase` | Break a messy commit into smaller, focused commits for interactive rebase. |\n| `/deconflict` | Resolve rebase/merge conflicts with clear explanation. |\n\n### Clipboard Utilities\n\n| Command | Description |\n|---|---|\n| `/clipdoc` | Copy file contents to the system clipboard. |\n| `/clippath` | Copy a file's absolute path to the system clipboard. |\n\n> **Attribution:**\n> - `/design-and-plan` inspired by [obra/superpowers](https://github.com/obra/superpowers/) brainstorming and writing-plans skills.\n> - Planning suite, git workflow, code quality, and clipboard commands adapted from [jmchilton/claude-jmchilton-plugins](https://github.com/jmchilton/claude-jmchilton-plugins) and [jmchilton/claude-commands](https://github.com/jmchilton/claude-commands).\n\n## Hooks\n\nHooks live in `hooks/` and are symlinked to `~/.claude/hooks/`. Configured in `~/.claude/settings.json` with portable `~/` paths.\n\n```bash\n# Setup on a new machine\nmkdir -p ~/.claude/hooks\nln -s $CLAUDE_METADATA/hooks/safety ~/.claude/hooks/safety\nln -s $CLAUDE_METADATA/hooks/peon-ping ~/.claude/hooks/peon-ping\n```\n\n### Safety Hooks (`hooks/safety/`)\n\n#### Blocking (PreToolUse)\n\n| Hook | Matcher | What it does |\n|---|---|---|\n| `git-guard.sh` | `Bash` | Blocks `reset --hard`, `push --force`, `clean -f`, `checkout -- .`, `stash drop/clear`, `branch -D` |\n| `protect-files.sh` | `Write\\|Edit` | Blocks edits to `.env`, `raw/`, `datasets/`, `.log`, lock files |\n\n#### Context Preservation\n\n| Hook | Event | What it does |\n|---|---|---|\n| `context-reinject.sh` | `SessionStart` / `compact` | Re-injects branch, commits, conda env, project-config, and PROGRESS.md after compaction |\n| `transcript-backup.sh` | `PreCompact` | Saves transcript before compaction (keeps last 20) |\n\n#### Progress Tracking\n\n| Hook | Event | What it does |\n|---|---|---|\n| Prompt hook | `SessionEnd` (`clear\\|logout\\|prompt_input_exit`) | Claude writes/updates PROGRESS.md with task summary, accomplishments, next steps, key decisions. Also updates MANIFEST.md if present. |\n| Prompt hook | `PreCompact` | Same as above — preserves context before compaction |\n| `progress-update-fallback.sh` | `SessionEnd`, `PreCompact` | Fallback if prompt hook fails — writes mechanical PROGRESS.md from git state (branch, recent commits, changed files) or filesystem (recent files, project contents). Skips if PROGRESS.md was updated in the last 60s. |\n\n#### Auto-formatting (PostToolUse, async)\n\n| Hook | Matcher | What it does |\n|---|---|---|\n| `auto-format.sh` | `Edit\\|Write` | Runs `ruff format` on .py files |\n| `notebook-strip.sh` | `Edit\\|Write\\|NotebookEdit` | Runs `nbstripout` on .ipynb files |\n| `command-audit.sh` | `Bash` | Logs commands to `.claude/command-audit.log` |\n\n### Peon Ping (`hooks/peon-ping/`)\n\nSound notification system that plays audio cues on hook events (session start/end, tool use, errors, etc.). Supports multiple voice packs, volume control, and push notifications (ntfy, Pushover, Telegram) via environment variables.\n\n| Config | Description |\n|---|---|\n| `config.json` | Volume, active pack, rotation mode, notification settings |\n| `packs/` | Voice pack sound files |\n| `scripts/` | Helper scripts for renaming, usage tracking |\n\n## Adding New Skills\n\n```bash\nmkdir skills/category/my-skill\n# Create SKILL.md with frontmatter (name, description, version, allowed-tools)\n# Keep SKILL.md under 500 lines — move details to supporting files\n# Add to enable-skills.sh suggested list if appropriate\n```\n\nSee `skills/claude-meta/skill-management/` for the complete guide.\n",
  "bytes": 11583,
  "sha": "11710589bfcf96cf045cd58dd2121a739e505607c939edacf8c20dbdea6c7cd1",
  "repo_slug": "delphine-l/claude_global",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/skl_delphine_l_claude_global_token_efficienc_705c7985/readme"
}